Kr210 kinematics update

[jwhendy]

This surely still needs work, but is a first stab at #98 . Creating the PR so we can talk code instead of theory and pictures! This addresses the kinematics of the original L150 and also attempts to bring in the L150-2, which is the same physical model with different limits/zero position defined for joint_a3.

[gavanderhoorn]

This looks like a nice cleanup. Thanks for that @jwhendy. :+1:

I'm still a bit foggy on why that offset is needed, but I think I'll just have to re-read your previous explanation(s) to figure that out (again).

I think @BrettHemes was more involved here, so perhaps he could also take a look?

[jwhendy]

@gavanderhoorn I think this is the simplest I've managed to try and explain it:

Here's one more shot... honestly the time between looking at this has me needing to re-remember as well :) It's down to a balance of three factors:

• re-using the same CAD for both robots
• adhering to the Kuka manual convention for joint limits
• only one non-zero offset between joints

Basically, you can pick any two of the above but you can't have all three. I was under the assumption that re-using the CAD was a definite to include, and all of these questions have been around which to give up among the second two.

• add an additional pi/2 offset to joint_a3 on the L150 in order to preserve the -219/+65 convention
• don't use this offset, but define the L150's limits to be -119/+155 like the L150-2.

Lastly, I admit to being hung up on trying something clever like your suggestion on #98. At this point, I think the PR is held up on me figuring that out in order to avoid a separate, mostly redundant .xacro.

I think I'll just create a separate .xacro later tonight and update this. If down the road someone wants to merge them in some clever way, they always can. Thoughts on that?

---

Edit: also, the guidance from @BrettHemes steered me toward adding the offset and preserving the manual convention, so that's what I did here.

The part of this PR that needs review was what folks thought of the probably stupid attempt shown here at adding a parameter to specify the L150 vs. L150-2 without a separate .xacro. This would require documentation that the user should pass a value for a model param if they want to use the L150-2.

[jwhendy]

I just split the L150 and L150-2 into separate .xacro, _macro.xacro, and test/*.launch files.

• same CAD
• preserves Kuka joint limits
• L150-2 has an additional pi/2 rotation at joint_a3

If you liked the model param... I think you can pull the last commit instead of the most recent.

One uncertainty: are dashes forbidden in file names? I have -2 identifying the L150-2 from the L150. Not sure how much of a foul that is; I can update to _2 if that's preferred.

[jwhendy]

Okay, after fiddling with the usage of an arg on the lbr_iiwa and thinking that was elegant for the flange, I went back to it here, too. We now have:

• reusable CAD
• an arg, variant, that defaults to "" for th KR210 L150 and can be specified as "_2" for the KR210 L150-2
• the variant arg simply changes the offset for joint_a3 and it's limits so that we match the Kuka manual definitions for each

I think that's pretty reasonable and a nice compromise (all we give up is one additional non-zero joint offset in the case of the L150). Good to go?

[BrettHemes]

As with the others, (#117 and #118) I like this approach. When I get some time I will load them up and check them unless @gavanderhoorn gets to it first. Hopefully this weekend.

[jwhendy]

I'll look at the reviews more closely tonight; much appreciated. One main question... I see comments on files with -2, which I deleted/moved away from. Do you want those back? With the approach I took on the most recent commit, I assumed we would treat these identically, minus passing _2 to trigger a different zero reference and limits for joint_a3.

I didn't plan to have a _2 namespace via the launch if you're using the L150-2 either... do we need to differentiate these variants via a namespace?

I can look into moving the code into the main _macro.xacro.

[BrettHemes]

which I deleted/moved away from

The load/test files for the -2 are still there and as it stands these are broken. If you delete them you can ignore the respective comments.

[jwhendy]

The load/test files for the -2 are still there and as it stands these are broken.

I missed that last comment before reading all the reviews. Yes, I intended to do completely away with the -2 via any files. I think I goofed and never did git rm... fixed now.

Thanks for the insight to just move top level; I agree for a simple if check, this is much cleaner.

[BrettHemes]

I was going to comment on removing the mass/inertia values but they all seem wrong (this was also pointed out in #36 it seems but never fixed). I am perfectly fine getting rid of these now and then adding them back later when good approximates are figured out (see discussion in #125)

[storeyp]

Hi folks, New to this world, so apologies if jumping in at the wrong place. I'm looking to use ROS on our two Kuka KR210 robots. The spanner in the works: they're the variant KR210 R2700, and seeing the length of discussion around L150 and L150-2 variants, I'm wondering if this will be problematic! I've connected successfully with the robots, and getting data back and forth, but it does seem that I've got problems with axis definition - the real world doesn't match what I see in rviz.

Any further changes since May 2018? What's the best way for me to proceed?

Thanks!

[jwhendy]

Sorry, I'm not sure how to answer. I believe for this, #117 and #118 just need time from @gavanderhoorn . I have not heard anything on these PRs since May 2018, minus similar inquiries.